Temporal dynamics and spatial variability in the enhancement of canopy leaf area under elevated atmospheric CO 2

Temporal dynamics and spatial variability in theenhancement of canopy leaf area under elevatedatmospheric CO2

H E A T H E R R . M C C A R T H Y *, R A M O R E N *, A D R I E N C . F I N Z I w , D AV I D S . E L L S W O R T H z,H Y U N - S E O K K I M *, K U R T H . J O H N S E N § and B O N N I E M I L L A R }*Nicholas School of the Environment and Earth Sciences, Duke University, Box 90328, Durham, NC 27708, USA, wDepartment of

Biology, Boston University, 5 Cummington Street, Boston, MA 02215, USA, zCentre for Plant and Food Science, University of

Western Sydney, Locked Bag 1797, Penrith South DC, NSW 1797, Australia, §Southern Research Station, USDA Forest Service,

3041 Cornwallis Road, Research Triangle Park, NC 27709, USA, }Carolina Mountain Land Conservancy, PO Box 2822,

Hendersonville, NC 28793, USA

Abstract

Increased canopy leaf area (L) may lead to higher forest productivity and alter processes

such as species dynamics and ecosystem mass and energy fluxes. Few CO2 enrichment

studies have been conducted in closed canopy forests and none have shown a sustained

enhancement of L. We reconstructed 8 years (1996–2003) of L at Duke’s Free Air CO2

Enrichment experiment to determine the effects of elevated atmospheric CO2 concentration

([CO2]) on L before and after canopy closure in a pine forest with a hardwood component,

focusing on interactions with temporal variation in water availability and spatial variation

in nitrogen (N) supply. The dynamics of L were reconstructed using data on leaf litterfall

mass and specific leaf area for hardwoods, and needle litterfall mass and specific leaf area

combined with needle elongation rates, and fascicle and shoot counts for pines. The

dynamics of pine L production and senescence were unaffected by elevated [CO2], although

L senescence for hardwoods was slowed. Elevated [CO2] enhanced pine L and the total

canopy L (combined pine and hardwood species; Po0.050); on average, enhancement

following canopy closure was �16% and 14% respectively. However, variation in pine Land its response to elevated [CO2] was not random. Each year pine L under ambient and

elevated [CO2] was spatially correlated to the variability in site nitrogen availability (e.g.

r2 5 0.94 and 0.87 in 2001, when L was highest before declining due to droughts and storms),

with the [CO2]-induced enhancement increasing with N (P 5 0.061). Incorporating data on

N beyond the range of native fertility, achieved through N fertilization, indicated that pine

L had reached the site maximum under elevated [CO2] where native N was highest. Thus

closed canopy pine forests may be able to increase leaf area under elevated [CO2] in

moderate fertility sites, but are unable to respond to [CO2] in both infertile sites

(insufficient resources) and sites having high levels of fertility (maximum utilization of

resources). The total canopy L, representing the combined L of pine and hardwood species,

was constant across the N gradient under both ambient and elevated [CO2], generating a

constant enhancement of canopy L. Thus, in mixed species stands, L of canopy hardwoods

which developed on lower fertility sites (�3 g N inputs m�2 yr�1) may be sufficiently

enhanced under elevated [CO2] to compensate for the lack of response in pine L, and

generate an appreciable response of total canopy L (�14%).

Keywords: broadleaf leaf area, drought, leaf area index, leaf area profile, Liquidambar styraciflua,

nitrogen availability, Pinus taeda

Received 4 May 2005; revised version received 16 November 2006 and accepted 15 May 2007

Correspondence: Heather R. McCarthy, Department of Earth

System Science, University of California, Irvine, CA 92697-3100,

USA, tel. 1949 824 2935, fax 1949 824 3874, e-mail:

[email protected]

Global Change Biology (2007) 13, 2479–2497, doi: 10.1111/j.1365-2486.2007.01455.x

r 2007 The AuthorsJournal compilation r 2007 Blackwell Publishing Ltd 2479

Introduction

Canopy leaf area index (area of leaves above a unit of

ground area; L) affects light interception and thereby

stand productivity (Jarvis & Leverenz, 1983; Waring,

1983; Vose & Allen, 1988). The L in the upper canopy

strata determines the light availability at lower strata,

and thus species dynamics in the subcanopy (Pearcy,

1990; Naumberg & Ellsworth, 2002; Augspurger &

Bartlett, 2003). Furthermore, L affects rainfall intercep-

tion and transpiration, and thus forest hydrology and

soil moisture (Stogsdill et al., 1989; Oren et al., 1998),

with impacts on soil respiration (Hanson et al., 1993;

Davidson et al., 1998; Palmroth et al., 2005).

Many forests do not reach their potential maximum

L because of water or nutrient limitations (Brix, 1981;

Linder, 1987; Vose & Allen, 1988; Albaugh et al., 1998). It

is expected that elevated atmospheric CO2 concentra-

tion ([CO2]) might alleviate certain limitations to L (e.g.

Woodward, 1990). Seedlings and saplings grown under

elevated [CO2] often show dramatic increases in leaf

area (Kellomaki & Wang, 1997; Tissue et al., 1997), but

this effect largely reflects accelerated ontogeny, in that

larger plants have more leaves (Pataki et al., 1998a;

Norby et al., 2003). These studies are unable to predict

how elevated [CO2] would affect L in a closed canopy

forest where maximum L may be limited by water,

nutrients or light.

The main mechanism by which elevated [CO2] could

alleviate water limitations to leaf area development

would be through reductions in stomatal conductance

leading to increased available soil water (e.g. Wood-

ward, 1990; Drake & Gonzalez-Meler, 1997). Overall,

observed reductions in conductance have averaged

around 20% (Drake & Gonzalez-Meler, 1997; Medlyn

et al., 2001), although Curtis & Wang (1998) showed a

nonsignificant 11% reduction. However, few studies

have shown significant reductions in conifer species.

Previous studies from the Duke Forest FACE site have

failed to show reductions in loblolly pine (Pinus taeda

L.) stomatal conductance with elevated [CO2] (Ells-

worth et al., 1995; Ellsworth, 1999), as have other studies

with loblolly pine (Lui & Teskey, 1995; Wang et al., 1995;

Murthy et al., 1996; Maier et al., 2002). Unchanging

stand-level water-use under elevated [CO2] at the Duke

FACE is consistent with these findings (Schafer et al.,

2002). Elevated [CO2] could also help forests overcome

nutrient limitations, if increased photosynthetic effi-

ciency leads to increases in nitrogen use efficiency (i.e.

less nitrogen being bound in the photosynthetic appa-

ratus means more nitrogen for plant growth). While

several studies find reductions in N concentrations on a

mass basis (Murthy et al., 1996; Curtis & Wang, 1998),

few definitively find reductions on an area basis (Me-

dlyn et al., 1999; Maier et al., 2002; Rogers & Ellsworth,

2002).

Another proposed mechanism for deeper canopies

and higher L with elevated CO2 is the lower light

compensation points (light level at which photosynth-

esis equals respiration) often predicted under elevated

[CO2] (e.g. Drake & Gonzalez-Meler, 1997). Although

some studies have found elevated [CO2] to lower the

light compensation point (Wang et al., 1995; Maier et al.,

2002), others have found reductions only in the upper

canopy where it is not likely to result in increased

L (Norby et al., 2003). Lower light compensation points

have not been observed under elevated [CO2] for lo-

blolly pine in the Duke FACE experiment (as reported

in Schafer et al., 2003). Furthermore, Maier et al. (2002)

showed a decrease in shoot needle density, (and thus

clumping) in loblolly pine trees under elevated [CO2]

which would decrease light penetration through the

canopy (Stenberg, 1996a; Stenberg et al., 2001; Palmroth

et al., 2002), potentially negating any positive influence

of elevated [CO2] on light compensation.

In addition, some studies have found [CO2] effects on

L dynamics, particularly changes in leaf senescence

patterns (Jach & Ceulmans, 1999; Li et al., 2000; Sigurds-

son, 2001; Karnosky et al., 2003; Tricker et al., 2004). Such

changes could offset or accentuate [CO2] effects on

L production.

Presently, few studies in closed canopy forests have

assessed the response of L to [CO2] enrichment, and

none have shown any sustained or consistent effect

of elevated [CO2] on the magnitude or dynamics of

L (Hattenschwiler et al., 1997; Gielen et al., 2003; Norby

et al., 2003). These studies have been conducted in

deciduous forests and may not be indicative of the

potential for response in coniferous forests. Thus, the

objective of this study was to quantify the effect of

[CO2] on L in a loblolly pine plantation growing at a

site with substantial spatial variation in hardwood pre-

sence and N availability and temporal variation in water

availability.

In practice, estimating L in loblolly pine is difficult.

Loblolly pine – unlike many evergreen conifers – has

a rapid needle turnover rate (�19 months; Zhang &

Allen, 1996), resulting in highly dynamic seasonal L,

with the maximum L reaching as much as twice the

minimum L (Kinerson et al., 1974; Dougherty et al., 1994;

Vose et al., 1994). Commonly used methods cannot

generate accurate L estimates in such canopies where

more than one cohort of needles is present at most

times, yet a large proportion of L is replaced annually.

Estimates based on allometry do not capture the effects

of interannual variation in climate such as droughts or

wind and ice storms, and optical estimates frequently

have a bias, particularly in coniferous canopies (Gower

2480 H . R . M C C A R T H Y et al.

r 2007 The AuthorsJournal compilation r 2007 Blackwell Publishing Ltd, Global Change Biology, 13, 2479–2497

& Norman, 1991; Sampson & Allen, 1995; Stenberg,

1996b).

Our study, performed at the Duke Free Air CO2

Enrichment (FACE) experiment, used data on litterfall

to estimate leaf production. We combined leaf produc-

tion estimates with specific leaf area (SLA) values, and

for the pine with measurements of needle elongation,

and fascicle and shoot counts, to reconstruct an 8-year

record of seasonal dynamics of L (see overview of the

procedure in Fig. 1). This record of L estimates, span-

ning a broad range of climatic conditions and natural

disturbance events, was used to address hypotheses on

the effect of elevated [CO2] on L, and to provide input

for use in modeling carbon and water cycle processes.

Our objective was to examine the potential effects

of elevated [CO2] on pine and hardwood leaf area

dynamics, focusing on the following questions: Can

elevated [CO2] increase peak and mean L in a closed

canopy conifer forest? Does elevated [CO2] affect the

timing of leaf production or loss? Is spatial variation in

the [CO2]-induced enhancement of L related to nitrogen

availability? Does elevated [CO2] change the response

of L to environmental stress? Based on results from

comparable studies in hardwood forests (Hattenschwi-

ler et al., 1997; Gielen et al., 2003; Norby et al., 2003), we

hypothesized that elevated [CO2] would have little

impact on stand L, either in terms of magnitude or

dynamics, once our forest canopy closed.

Materials and methods

Site description

The study site is within a loblolly pine (Pinus taeda L.)

plantation established in 1983 on moderately low ferti-

lity, acidic clay-loam (Enon Series) in Duke Forest in

Orange County, North Carolina (351580N, 791080W;

elevation 163 m). The climate is warm and humid in

summer and moderate in winter with a mean annual

temperature of 15.8 1C. Precipitation is distributed ap-

proximately evenly throughout the year, with a 111-year

average of 1145 mm. In early 2002 pines reached

�18 m in height and made up 90% of the basal area.

Common broadleaf species include sweetgum (Liqui-

dambar styraciflua) in the mid to upper canopy, and Acer

rubrum, Ulmus alata and Cornus florida in the mid to

lower canopy.

In 1994, the FACE prototype (Plot 7), a 15 m radius

plot, and an adjacent untreated reference plot (Plot 8)

were established. CO2 enrichment (550 ppm during

Litter mass

Pine Hardwood

Litter massExpansion

function

SLA

Leaf longevity

Needle count Needle length

SLA Degree days

Shoot count

Vertical Lprofile

Lloss

Lloss

Lgain

Lprod

Lprod

Lexp

Lexp

Lrel

Lpeak

Lpeak

Lmin

L

L

L

L

Fig. 1 Overview of methodology for calculating leaf area and leaf area derived variables presented in this paper. Ovals indicate inputs

(measured inputs with solid borders, parameters with dashed borders) and diamonds represent measured input variables necessary

only under disturbance conditions (see ‘Materials and methods’). Squares are calculated variables, with annual values in thick borders,

and daily time steps in thin borders. Gray shading highlights the leaf area variables that will be the focus of analysis and discussion in

this paper. Lprod 5 leaf area production; Lexp 5 leaf area expansion; Lloss 5 leaf area loss; Lpeak 5 annual maximum leaf area; L 5 average

(annual or functional) leaf area index; Lmin 5 annual minimum leaf area index. SLA, specific leaf area.

VA R I AT I O N I N C A N O P Y L E A F A R E A R E S P O N S E T O C O 2 2481


daylight hours of the growing season) commenced in

1994 according to the FACE protocol (Hendrey et al.,

1999). The replicated FACE (Plots 1–6; also 15 m radius)

was subsequently established in 1996 when CO2 enrich-

ment (1 200 ppm) was initiated in three of the plots. In

1998, Plots 7 and 8 were each split in half, and one

half of each received yearly nitrogen fertilization

(11.2 g N m�2 yr�1) which was also applied through a

7.5 m buffer arcing outside the fertilized half. Four pairs

of auxiliary plots (10 m� 10 m with 10 m wide buffer)

were established nearby, and one member of each pair

was fertilized (Oren et al., 2001). The average actual CO2

concentration during 1996–2003 was 570 ppm, with

490% of 1-min [CO2] averages within 20% of the target

(average target 5 572 ppm). Figure 2 shows the timeline

for the application of [CO2] and fertilization treatments

to each experimental plot.

Canopy and leaf measurements

In this section we describe the measurements that

allowed the characterization of the seasonal and inter-

annual development of leaf area (refer also to Fig. 1).

Litterfall. For the six replicated FACE plots we used

litterfall data previously published only as annual

values (Finzi et al., 2002, 2006). Litter falling into 12

0.16 m2 baskets in each plot was collected bimonthly

during peak litterfall (September–December) and

monthly otherwise. Samples were dried (65 1C, 4

days), separated into pine needles or hardwood

leaves, and weighed. Beginning in fall 2001, each plot

within the FACE prototype complex (Plots 7 and 8 and

the auxiliary plots; see Fig. 2) was equipped with four

0.5 m2 wood frame litter baskets. The processing

protocol was as described above, but sweetgum leaves

were assigned an independent category. No differences

were observed between pine and hardwood foliage

mass collected with the two types of collectors

(P 5 0.600 and 0.888, respectively) during a 5-month

calibration period in which the collectors were

colocated in all plots.

Specific leaf area. In the FACE prototype complex, from

each autumn collection (September through December,

2001–2003), before litter drying, five needles/leaves

each of pine, sweetgum, and all other hardwoods

combined were used to determine SLA (cm2 g�1).

Projected leaf area was measured optically (DIAS,

Decagon Devices Inc., Pullman, WA, USA), samples

were oven dried (70 1C; �3 days) and weighed.

Although previous work at this site did not find a

significant difference between litter and green leaf/

needle SLA (Finzi et al., 2002), SLA did differ between

senesced and green needles collected after the 2002 ice

storm (P 5 0.003), and therefore green needle SLA was

employed for green needles dislodged in that event.

None of the species groups showed any trend in litter

SLA within a year, or any significant differences

between treatments (all P40.050). Therefore, litter

SLA values were averaged within years and across

treatments. For pine, litter SLA values measured in

2001 were used for conversions of leaf litterfall to leaf

area for all years before 2001.

To account for the effect of light environment on SLA

of sweetgum and other hardwoods mostly present in

the lower and subcanopy, SLA from 2001 was reduced

by multiplying by the ratio of pine L at time of

hardwood bud-break in each year before 2001 and

pine L in 2001.

Optical measurements. In order to contrast our estimates

with values from a known, albeit biased, method,

optical gap fraction measurements were taken using a

LAI-2000 canopy analyzer (Li-Cor, Lincoln, NE, USA).

In the FACE prototype complex, measurements were

taken at four locations per plot, every 1–2 months

during the periods of August 1999–July 2000 and

October 2001–December 2003. In the replicated FACE

plots, measurements were taken every 1–2 months at

eight locations per ring between August 2002 and

December 2003. Viewcaps (901 for FACE prototype

complex and 1801 for replicated FACE) were used for

all measurements.

The LAI-2000 was also used to generate vertical L

profiles (leaf area density at 2 m resolution; LAD) for the

pine canopy during periods in which hardwoods were

Plot

1 15 1.23.15.23.95.23.9

3.6 (9.2)4.1 (9.7)

4.2*9.5*

15151515151510

2345678

aux cont (× 4) 10 × 1010 × 10aux fert (× 4)

Size(m)

Available N(g m ) Experimental period

1994 1996 1998Year

2000 2002

Fig. 2 Overview of experimental design and treatment regime

throughout the experimental period. Plot size is given as either

plot radius or dimensions. Auxiliary control and fertilized plots

are comprised of four plots each, with the available nitrogen

values given representing the average of the four plots. Available

nitrogen values in parenthesis are for the fertilized halves of the

plots, assuming half of the applied fertilizer is available. Gray

shading indicates the period of time during which a subset of the

plots were exposed to treatment, but not measured.



leafless. These measurements were taken in four

cardinal directions on four dates (16 November 2002;

7 March 2003; 26 November 2003 and 12 March 2004) in

the replicated FACE and the FACE prototype plots from

the central walkup towers, and on two dates (7 March

2003 and 12 March 2004) in the reference and auxiliary

plots.

In March of 2002, crown length for all pines in the

FACE prototype complex plots, and within a subplot of

each replicated FACE plot (1.5 m extending from plot

boardwalks) was determined as the difference between

total tree height and height to crown base measured

with a survey laser (Criterion 400, Laser Technology

Inc., Englewood, CO, USA).

Elongation of needles. Needle elongation in the FACE

prototype complex was quantified in 2002, and in the

replicated FACE in 1998 and 1999 (Rogers & Ellsworth,

2002). Needles at lower, middle and upper canopy

positions of three trees were monitored in each plot.

Needle length was obtained at 1–3 week intervals on all

current year flushes. The trajectory of relative expansion

of new foliage in each third of the canopy was scaled to

relative leaf area expansion (Lexpp) for the pine canopy

by weighting the trajectories according to the

proportion of leaf area in each canopy third. For years

in which elongation measurements were not conducted,

the relative growth trajectory from the year that was

most climatically similar (in terms of precipitation and

temperature) was employed.

Needle and shoot counts. Between mid-July and mid-

September 2002, fascicles were counted monthly on

each shoot measured for needle elongation. Counts

were made of needles in all flushes of 2002, and the

last flush of 2001. The fraction of 2002 needles per shoot

(Nf02) at each measurement time was related to the

vertical position within the canopy. In November of

2004, counts of current and previous year flushes

were made at 2 m intervals through the crowns of

trees (three to seven trees) of five plots spanning the

[CO2]� fertilization treatments. The fraction of shoots

with current year flushes (Sfc) was regressed against the

relative height within the canopy.

Base of live crown sapwood area. In April 2002, diameter at

breast height (dbh, cm) and at the base of the live crown

(dblc; cm) was measured on 48 trees in plots of the

FACE prototype complex and replicated FACE, using

towers and canopy lifts. These measurements were

used to develop a regression between dblc and dbh

(r2 5 0.87, Po0.001). This regression, dblc 5 0.74�(dbh)�16.40� (stand density) 1 0.18, where stand

density is trees per m2, was then used to calculate the

diameter at the base of live crown of all pines within the

replicated FACE and FACE prototype complex plots.

Sapwood area was taken to be 85% of base of live crown

total cross sectional area (Schafer et al., 2002).

Calculation of leaf area index (1996–2003)

L at any point in time is the balance between growth

and loss of foliage, and requires accounting for both

dynamics and magnitude. Due to differences in foliage

dynamics, procedures for characterizing pine and hard-

wood leaf area index will be described separately,

beginning with pine (refer also to Fig. 1).

Pine. Annual leaf area production of pine (Lprodp;

m2 m�2) was derived as

Lprodp ¼ LM� SLA; ð1Þ

where LM is annual pine needle litter mass (g) (Table 1).

For loblolly pine, the process of equating needlefall

mass to yearly production is complicated by an

average needle longevity of 19 months (Zhang &

Allen, 1996), such that most of the needlefall within a

given year represents the needles produced during the

previous year. To estimate Lprodp from the timing and

amount of leaf area loss (Llossp), we calculated the

annual Lprodp for a given year as the sum of litterfall L

collected during the period of time spanning May of the

Table 1 Frequently used symbols

Definition Units

Subscripts

c Canopy (pine 1 hardwood)

h Hardwood

i Canopy layer (i 5 2 m interval or

canopy third)

p Pine

t Time (day of year)

Symbols

As Sapwood area m2 m�2

L Leaf area index m2 m�2

Lexp Relative leaf expansion

Lgain New leaf area gain m2 m�2

LM Annual litter mass of leaves or needles g

Lloss Leaf area loss m2 m�2

Lmin Annual minimum leaf area m2 m�2

Lpeak Annual maximum leaf area m2 m�2

Lprod Leaf area production m2 m�2

Lrel Leaf area relative to annual maximum

N Soil nitrogen g m�2

RWC Relative water content

SLA Specific leaf area cm2 g�1

y Volumetric soil moisture m3 m�3



following year, until May of the year after that. For

example, the production of 1998 would be calculated as

the needle mass falling from May 1999 through April

2000.

A continuous trajectory of absolute leaf area gain was

generated by

LgainðtÞ ¼ Lprodp � LexppðtÞ; ð2Þ

where t is day of year and Lexpp is relative leaf area

expansion from needle elongation measurements. Inter-

polation of litterfall L between collection times produced a

continuous trajectory of Llossp. The trajectories of Lexpp and

Llossp were combined at a daily time step to produce a

continuous record of L:

LpðtÞ ¼ Lpðt�1Þ þ LgainðtÞ � LlosspðtÞ: ð3Þ

This approach to estimating Lprodp [Eqn (3)] was

successful over large periods of this study. However, the

assumption that the timing of leaf production can be

determined based on leaf loss is invalid when

disturbance events cause premature loss of needles.

Between 2001 and 2004 the site experienced four such

perturbations: a moderate late summer drought in 2001, a

severe spring to late summer drought in 2002, a severe ice

storm in December 2002 and a moderate hurricane in

September 2003. These events decoupled the timing of

Llossp from Lprodp because needles could be lost during the

same year in which they were produced.

Most difficult to resolve was the partitioning of foliage

lost during 2002, due to the effect of both the severe

drought and ice storm. We dealt with the cumulative

effects of these losses by using information about the age

distribution of needles before current year needles were

affected by the drought (day 193). Determining the

proportion of L produced in 2001 and 2002 at that time

allowed for the calculation of Lprodp in 2002, with the

balance of L representing needles remaining from 2001.

The partition of L between 2001 and 2002 foliage was

determined for each canopy third (i 5 1–3) by multiplying

the fraction of 2002 needles (Nf02) with the fraction of

current year shoots (Sfc) and pine leaf area found in that

canopy third (Lf):

Lf02ðiÞ ¼ Nf02ðiÞ � SfcðiÞ � LfðiÞ; ð4Þ

where the fractions of 2002 needles and shoots were

determined from needle and shoot counts, and Lf was

determined from the vertical optical measurements. Lprodp

in 2002 was then calculated by summing Lf02 from each

canopy third and multiplying by pine L at day 193:

Lprodp ¼ Lpð193Þ � Lf02: ð5Þ

We adjusted L(193) to reflect the L which would have

been present had 2002 needles been fully expanded.

To calculate the amount of 2001 foliage lost during the

2001 drought, we assumed that the proportion of foliage

lost prematurely was half of that lost prematurely during

the drought of 2002. For the green leaf mass lost during

Hurricane Isabel in September 2003, we assumed that the

loss of leaf area from each cohort (2002 and 2003) was

proportional to the quantity of each cohort in the canopy,

reflecting an equal probability of mechanical dislodging.

Hardwood. Because there is no significant component

of marcescent hardwood species (with dead leaves

persisting on trees) at the site, annual foliage

production was based on annual litterfall with no

time lag. We did not quantify the temporal dynamics

of leaf expansion, so hardwood leaf growth dynamics

were calculated based on several parameters. Firstly,

degree-day sums were used to determine the day at

which leaf expansion began. Budbreak for hardwood

species generally coincided with the beginning of pine

diameter growth, for which 5 years of observations

determined to begin at 426 degree-days (SD � 21

degree-days). Degree-days represented the summation

of mean daily temperature (minus a base temperature

of 2 1C) beginning after the last period of 4 or more

sequential days with mean daily temperature in each

day of �2 1C. Secondly, based on a literature survey

(Hunter & Lechowicz, 1992; Augspurger & Bartlett,

2003), 30 days was selected as the time necessary to

complete leaf expansion, with relative leaf expansion

(Lexph) occurring exponentially as in Oren & Pataki

(2001). This assumes that after the initial pulse of leaf

expansion, indeterminant species produced relatively

little additional foliage. The relative pattern of leaf loss

was determined according to litterfall measurements. A

small quantity of leaves occasionally remained attached

long after senescence, resulting in a small amount of

leaf fall through the winter. The date by which 95% of

total litterfall, summed between two consecutive bud

breaks, occurred was determined to be the date of full

senescence; leaf fall occurring later was assigned to this

date. At the daily time scale, relative L (Lrel) was

determined by

Lrel ¼ Lexph � Llossh; ð7Þ

where Llossh is the relative loss of foliage. Finally,

LhðtÞ ¼ LM� SLA� Lrel; ð8Þ

where LM is annual hardwood leaf litter mass (g).

Soil water content

Within the FACE prototype and reference plots, volu-

metric soil water content (y; m3 m�3) was measured

continuously from 2001 at eight locations, with four



probes at 30 cm and four probes at 10 cm (ThetaProbe

ML1x or ML2x; Delta-T Devices, Cambridge, UK).

Beginning in 1997, in each of the six replicated FACE

plots, y of the upper 30 cm was measured continuously

in four locations (CS615; Campbell Scientific, Logan,

UT, USA). All sensors were sampled every 30 s, and 30-

min averages were logged (21X or CR23X, Campbell

Scientific, Logan, UT, USA). Growing season averaged

relative water content (RWC) was calculated by divid-

ing growing season averaged y by the field capacity yat the site (0.54; Oren et al., 1998).

N availability

Annual net N mineralization rates (g N m�2 yr�1; Nmin)

for 1998 from the six replicated FACE plots were taken

from Finzi et al. (2002). The relationship between these

mineralization rates and leaf N concentrations was

applied to 1998 leaf N concentrations in the FACE

prototype complex to estimate N mineralization rates

in these plots. Available N was defined as

Navl ¼ Nmin þNdep þNfix þNfert; ð9Þ

where Ndep is N deposition (from Finzi et al., 2006) and

Nfix is N fixation (from Hofmockel & Schlesinger, 2007).

These studies show that losses of N from this system are

small. For Nfert, we assumed that half of the applied

11.2 g N m�2 yr�1 was available for plant uptake (Ducey

& Allen, 2001).

Statistical considerations

The Duke Forest FACE experiment includes the six

plots of the ‘replicated’ portion of the experiment

(n 5 3), and the FACE prototype complex. The FACE

prototype complex (Plots 7, 8 and auxiliary plots)

included five ambient plots receiving no fertilization,

which together with the three FACE plots gave a total of

eight ambient replicates (Fig. 2). The complex also

included five fertilized replicates. Given their concen-

tration in the southern end of the site and smaller size,

and in order to be conservative in interpreting the

results, we treated the ambient plots of the complex

collectively as one plot, resulting in four replicates each

for ambient and elevated [CO2] conditions, and one

each for fertilized and elevated [CO2]� fertilized con-

ditions. Thus, reduced to n 5 1, the fertilization treat-

ments were used to generate information on the upper

limits of leaf area under ambient and elevated [CO2]

conditions. Statistical differences between the unrepli-

cated (fertilized) and the replicated (unfertilized) treat-

ments were assessed with one-sample t-tests (Sokal &

Rohlf, 1995). The unfertilized treatments were first

investigated as to whether the prototype complex

(n 5 1) is a part of the ‘replicated’ population (n 5 3)

for each variable. The smallest P-value was 0.136, and

all other P-values were greater than 0.42, strongly

rejecting the hypothesis that the prototype and ‘repli-

cated’ plots are not from the same experiment. The

effects of elevated [CO2] on peak L, minimum L, aver-

age L, L production, ratio of annual maximum to mini-

mum L, and L-to-sapwood area ratio were analyzed

through repeated measures ANOVA with plots blocked

according to the pairing of plots established at the onset

of the experiment (n 5 4 blocks). [CO2] effects during

each year of the experiment were tested through linear

contrasts within the repeated measures ANOVA. Effects

of elevated [CO2] on leaf expansion and loss were also

tested through repeated measures ANOVA on the para-

meters of sigmoidal regressions of the form: a/

(1 1 exp�((x�x0)/b)). The relationship between functional

L and available N were evaluated using ordinary least-

square regression. These statistical analyses were con-

ducted in SAS (Version 8.0, Cary, NC, USA).

Results

Foliage production and expansion

In pines, maximum needle length varied through the

canopy, with lower canopy needles reaching only

�50% of upper canopy needle length (Fig. 3a). Repeated

measures analysis of the parameters of sigmoidal elonga-

tion trajectories – developed by weighting the trajectories

of each canopy third by its leaf area – indicated that

expansion patterns were not different between ambient

and elevated [CO2] during any of the 3 years measured

(Table 2). Effects of interannual variation in climate were

more pronounced than [CO2] effects, with drought years

showing a prolonged expansion period.

Combined for all years and both [CO2] treatments

during the period 1998–2002 (thus excluding the post-

ice storm year 2003), treatment-level Lprodp increased

with increasing minimum L (Lminp) – representing L at

the start of the growing season – and treatment-level

growing season average relative soil water content,

RWC. A multivariate least-square regression analysis

(SAS Proc Reg and Proc Stepwise) revealed that the

interaction of Lminp � RWC explained most of the varia-

bility in Lprodp under both [CO2] treatments (Table 3).

Under ambient [CO2], the parameter for the effect of

Lminp on Lprodp was significantly smaller, resulting in

lower leaf area production for a given set of conditions.

Foliage loss

Repeated measures ANOVA of regression parameters for

the relative pattern of Lloss of the pine and hardwood



species indicated some differences between years, as

well as treatments. For the pine year was a significant

effect (Po0.001), but the pattern of needle loss was

unaffected by [CO2] (P40.206; Table 2). Considering

the effect of high fertility (fertilization) through single

observation vs. sample mean t-tests (Sokal & Rohlf,

1995) indicated a more rapid loss of foliage during

drought under both ambient (2001 and 2002; Po0.033)

and elevated [CO2] (2002 only; P 5 0.031; Fig. 4c). There

was also year-to-year variation in loss patterns for

hardwood leaves (Po0.001). Elevated [CO2] slowed

Llossh relative to ambient [CO2] from 1999 to 2003 (all

Po0.005; Fig. 4d), also with indication of a more rapid

loss in fertilized plots during drought years. (Po0.05

in two of four tests).

For the pine, the maximum rate of Llossp (the derivative

of the inflection point of fitted sigmoidal curves) was

explained by RWC. Excluding data from 2001, in which

loss was enhanced by a drought beginning late in the

growing season, RWC explained 94% (Po0.001) of the

variation in Llossp with no significant difference in the

relationship under elevated and ambient [CO2] (Table 3).

Leaf area index

Over the years after L reached a quasi-equilibrium

(1999), pine L averaged 2.51 (�0.13) m2 m�2 under

ambient [CO2], and 2.90 (�0.14) m2 m�2 under

elevated [CO2]. During the same period, canopy

Fig. 3 Relative needle length as a function of canopy position

for pine under ambient CO2, elevated CO2, fertilization and

elevated CO2 with fertilization, during the 2002 growing season

(a), and canopy averaged relative needle expansion (Lexpp) for

pine over the course of the growing season for 1998, 1999 and

2002 growing seasons (b). The gray symbols in (a) show the

average values for each canopy third measured in the FACE site

in 1998 and 1999. Open symbols are used for ambient CO2 and

closed symbols are used for elevated CO2.

Table 2 Significance (P-values) of repeated measures ANOVA

factors on sigmoidal parameters for canopy averaged relative

needle expansion (Lexpp) for pine over the course of the

growing season for 1998, 1999 and 2002 growing seasons,

and average annually relativized leaf area loss (Lloss) for pines

and hardwoods (1998–2002)

Pine needle

expansion Pine needle loss

Hardwood leaf

loss

b x0 b x0 b x0

CO2 0.184 0.961 0.206 0.734 0.642 0.004

Year 0.165 0.016 o0.001 o0.001 o0.001 o0.001

CO2� year 0.198 0.250 0.755 0.088 0.653 0.009

Table 3 Coefficients for response functions of 1998–2002

pine leaf area production (Lprodp) 5 minimum L 1 (minimum

L� RWC) and rate of loss of pine leaf area 5 intercept 1

Lpeakp 1 RWC, where RWC is relative soil water content

Parameter Ambient Elevated

Production

Intercept 1.06

Lminp �0.08

Lminp � RWC 1.01

r2 0.55 0.98

Rate of loss

Intercept �0.27

RWC 1.72

r2 0.98 0.94

The model for pine leaf area production had an adjusted

r2 5 0.90 with both [CO2] treatments included, and for pine

leaf area loss, excluding data from 2001, the full model had an

adjusted r2 5 0.94. Only parameters that contribute signifi-

cantly to a reduction in unexplained variation are included,

and different values for the same parameter indicate difference

at Po0.05.



L averaged 3.34 (�0.17) m2 m�2 and 3.90 (�0.19)

m2 m�2 under ambient and elevated [CO2], respec-

tively. Overall, elevated [CO2] increased Lpeakp, Lprodp,

and the annual mean Lp, ðLpÞ (Appendix A). Elevated

[CO2] increased Lpeakp and Lp every year beginning in

1998 (Po0.050), and Lprodp during 5 of the 7 years since

1997 (Po0.050). While elevated [CO2] generally in-

creased all measures of L, the absolute effect was

dynamic from year to year (see Appendix A). The upper

limits of leaf area response were assessed using a single

observation vs. sample mean t-test (Sokal & Rohlf,

1995). According to this test, fertilization did not sig-

nificantly increase annual Lp (all P40.050). Similarly,

fertilization combined with elevated [CO2] did not

produce higher annual Lp than elevated [CO2] alone

(all P40.050).

For the hardwood, we found no significant effect

of any treatments on Lpeakh, Lprodh or annual Lh (all

P40.050). The lack of Lh response, and the nonuniform

distribution of hardwoods among plots, resulted in a

slightly smaller effect of elevated [CO2] when annual

mean canopy leaf area (pine 1 hardwood; Lc) was as-

sessed in comparison with results for the pine Lp (see

Appendix B). Nevertheless, elevated [CO2] significantly

increased annual Lc in every year beginning in 1998

(Po0.050). Assessments of the upper limits of leaf

area response at high fertility levels using a single

observation vs. sample mean t-test (Sampson & Allen,

1995) showed that fertilization increased Lc in 2001, but

fertilization under elevated [CO2] did not produce

higher annual Lc than elevated [CO2] alone (P40.050).

The reconstructed L over the entire span of the study

period (1994–2003) is shown in Fig. 4. The effects

of several hurricanes (Hurricane Fran in 1996, Floyd

in 1999; and Isabel in 2003) and a severe ice storm

(December 2002) are detectable in the L time series.

Diagnostics

In order to assess the reliability of our estimated pine

L, we considered three measures: (1) the relationship

between L determined from the combination of litter-

fall, SLA and supporting measurements and the L

determined optically, from LAI-2000, (2) the ratio of

annual maximum L to winter minimum L and (3) the

winter minimum leaf-to-sapwood area ratio. Firstly, Lc,

composed only of Lp during the leafless period for

hardwoods, was compared with that obtained from

optical measurements, after correcting the optical mea-

surements for clumping of pine foliage, and for woody

surface area (Stenberg, 1996b; Pataki et al., 1998b; There-

zien et al., 2007). After these corrections, optical esti-

mates of L had a more restricted range than the range

generated based on our method; it was greater at the

low end of L ( 1 0.4 m2 m�2) and lesser at the high end

(�0.3 m2 m�2). Thus, the average of the optical estimates

over the entire L range was similar to that estimated

based on our method ( 1 0.06 m2 m�2) regardless of the

Fig. 4 Average annually relativized leaf area loss (Lloss) under ambient CO2, elevated CO2, fertilization and elevated CO2 in

combination with fertilization for pines and hardwoods under nondrought (1999, 2000) (a, b) and drought conditions (1998, 2001,

2002) (c, d).



treatment. This matches the bias reported in other

comparisons performed in pine stands (Sampson &

Allen, 1995).

Secondly, under ambient [CO2] the mean ratio of max-

imum to winter minimum Lp for the entire study period

is close to 1.8, with large excursions above and below this

value occurring as a result of drastic climate-induced leaf

area reductions and subsequent recoveries (Fig. 6a). A

value of �1.8 is generated by needle longevity once a

loblolly pine canopy reaches steady state (Kinerson et al.,

1974). The fertilized treatments showed a greater max-

imum/minimum amplitude during the severe drought

year of 2002, reflecting greater sensitivity to drought

(Linder, 1987; Raison et al., 1992). Elevated [CO2] resulted

in a higher maximum-to-minimum L ratio in the first full

year of the [CO2] enrichment (1997; P 5 0.021), but other-

wise did not affect this ratio (P 5 0.104).

Thirdly, we calculated winter Lminp-to-sapwood area

ratio. Except in the last 2 years of the experiment, when

drought and the ice storm reduced the ratio, Lminp/Asp

under ambient [CO2] was similar to a locally generated

allometric value (Pataki et al., 1998b; Fig. 5b). Elevated

[CO2] had no effect on this ratio (P 5 0.590, all yearly

contrasts P40.05). Compared with elevated [CO2] alone,

adding fertilization resulted in a significantly higher

Lminp/Asp during 2002–2003 (Po0.050), but showed

greater sensitivity to drought through a more severe

reduction in Lminp/Asp during the last 2 years of the

study. Excluding the post-ice storm year 2003, fertiliza-

tion increased the annual Lminp/Asp by 18%, similar to

another study with loblolly pine (Ewers et al., 2000).

The peak Lp for the ambient plots – both fertilized and

unfertilized – based on the comprehensive method used

here, although high, is within the range reported for

loblolly pine (Vose & Allen, 1988; Hennessey et al., 2004;

Sword Sayer et al., 2004).

Leaf area enhancement

In order to account for the differential displays of pine

and hardwood L in the canopy L (i.e. hardwoods are

leafless part of the year), we evaluated the [CO2]-

induced enhancement of L during the period in which

foliage is active, representing the functional L. Functional

L, expressed as leaf area duration, is the L averaged over

months with mean temperature 49 1C for pine, and

over foliated months for hardwoods, multiplied by the

fraction of the year these months represent. The [CO2]-

induced enhancement ratio of the functional Lc was

generated by summing the functional L of pine and

hardwood. Using the pairing of plots established at the

onset of the FACE experiment, enhancement ratios for

each pair of plots were generated, and normalized

(divided) by their initial (pretreatment) ratio.

The [CO2]-induced enhancement ratio of the func-

tional Lp was significantly greater than zero beginning

in 1998, and averaged 16% (�1%; Fig. 7a). The elevated

[CO2] enhancement of functional Lc, averaging 14%

(�1%), was significant 5 of the last 6 years; it was

statistically insignificant in 2002 due to both decreasing

differences and increasing variability (Fig. 7b).

Although the effect cannot be tested directly, data from

the fertilized elevated [CO2] plot suggests that an upper

limit for [CO2] enhancement of functional Lp and Lc

is�30–40% and �30–50%, respectively.

Vertical leaf area distribution

The dates of vertical leaf area profile measurements

represent near peak and near minimum leaf area for

pines. The only point at which there was a significant

Fig. 5 Leaf area index from 1996–2003 for pine (a), hardwood

(b) and canopy (pine 1 hardwood) (c) with ambient CO2, ele-

vated CO2, fertilization and elevated CO2 with fertilization. Bars

indicate 1 SE, and are placed at the peak L of each year.

Hurricanes and changes in experimental configuration are noted.



effect of [CO2] was on day 320, 2002 (Fig. 8a), where the

level in which leaf area density was highest was shifted

upwards in the canopy under elevated [CO2] compared

with the canopy under ambient [CO2] (Po0.001). How-

ever, there were some changes in the profiles between

the time of maximum Lp (Fig. 8a and b) and minimum

Lp (Fig. 8c and d) for the treatments for which measure-

ments were available at both times. Compared with the

profile at minimum Lp, the profile at near maximum Lp

had a narrower relative canopy distribution, and the

relative position of the highest density of L was shifted

upwards in the profile under both ambient [CO2] and

elevated [CO2].

On an absolute basis, canopy length measurements

show that [CO2] lengthened tree crowns (after

accounting for stand density) by �0.3 m (P 5 0.004).

Fertilization did not have a detectable effect on crown

length (P 5 0.184).

Nitrogen effects on leaf area enhancement

Although the proportion of pine in the canopy at Lpeakc

decreased with decreasing N availability using, as an

index, the N availability estimates from 1998 (not

shown; r2 5 0.761, P 5 0.023), some compensation was

made through the response of hardwood L to [CO2]

(Fig. 9a vs. 9b). Here, we present the functional L, but all

other expressions of L depicted similar patterns. In 1996,

the first year of [CO2]-enrichment in the fully replicated

FACE experiment and before [CO2] induced an impact

on L, available N explained 89% of the variability in

functional Lp and 76% in Lc (P 5 0.005 and 0.023,

respectively), and there were no differences between

[CO2] treatments in the relationships of Lp or Lc to N

(P 5 0.695 and 0.073, respectively; Fig. 9a and b).

In 2001, the year in which L reached its maximum

(Fig. 5), the spatial variation in functional Lp was highly

correlated to N (r2 5 0.94 and 0.99 for ambient and

elevated [CO2]; P 5 0.161 and 0.051, respectively). More

importantly, the slope of the response of Lp to N was

steeper under elevated [CO2] (P 5 0.103 for Fig. 9a),

leading to an enhancement of Lp which increases with

N. At low N, there was no difference in Lp between the

[CO2] treatments. Despite the sensitivity of functional

Lp to available N, functional Lc under ambient and

elevated [CO2] was insensitive to N over the range of

N availability investigated (P 5 0.661 and 0.841, respec-

tively), but higher under elevated [CO2] (P 5 0.092).

This insensitivity to N leads to a constant enhancement

of functional Lc across the native N gradient (Fig. 9b).

N availability in the prototype complex was esti-

mated differently from that in the other six plots.

Nevertheless, adding the data from the unfertilized

plots of the prototype-complex to the data from the

other three plots in each treatment did not change the

relationship between functional Lp and available N

under either treatment (P 5 0.998 and 0.800 for ambient

and elevated [CO2], respectively; Fig. 9c). With more

statistical power (n 5 4 vs. n 5 3) the relationship under

elevated [CO2] was more clearly different from that

under ambient conditions (P 5 0.061). Similarly, the

relationships for functional Lc were not unaltered by

the addition of the unfertilized plots of the prototype

complex (Fig. 9d).

Discussion

Canopy leaf area index affects the recruitment and

growth of subcanopy individuals, and ecosystem pro-

cesses such as water, energy and carbon fluxes. We

evaluated the effects of elevated [CO2] on the magnitude

and dynamics of L in the canopy of a pine forest with a

variable hardwood component. Previous publications

from the Duke FACE site have suggested that [CO2]-

induced increases in L have been minimal (Lichter et al.,

2000; DeLucia et al., 2002). These assessments were

based on optical measurements or on a small subset of

the variables used in this study to reconstruct the leaf

area of pine, and thus could be subject to methodologi-

cal errors (optical measurements – Gower & Norman,

1991; Sampson & Allen, 1995; Stenberg, 1996b; Law et al.,

2001) or errors associated with unconstrained leaf area

dynamics and changes in allometry. The diagnostic

assessment of our L reconstruction demonstrated that

the results discussed below are well constrained.

We show that [CO2] had a significant impact on the

absolute values of the pine component of L, Lp (Fig. 5), yet

did not affect the seasonal dynamics of Lp (Figs 3, 4, and

6). Elevated [CO2] induced a 16% increase in functional

Lp, and a similar enhancement for the entire canopy, Lc

(14%; Fig. 7). However, the variation in the response

among plots within a treatment was not random – much

of the variation was determined by the spatial variation

in N availability and the associated change in the propor-

tion of the canopy L comprised of hardwood species.

Below we discuss the effects of [CO2] enrichment on the

temporal dynamics and the spatial average and distribu-

tion of L.

Treatment-average foliage expansion and loss

Evaluating the impact of elevated [CO2] on L requires

consideration of both leaf production and foliage devel-

opment and loss, because these dynamics determine

how long foliage is displayed and functions. While we

detected no systematic differences in the intra-annual

dynamics of pine L with elevated [CO2], either in terms

of development or loss, elevated [CO2] slowed leaf loss



for the composite of many hardwood species (Figs

3 and 4, Table 2). Other studies on the effect of [CO2]

on L dynamics have generated mixed results. While leaf

phenology did respond to elevated [CO2] in certain

species (Pinus sylvestris, Quercus myrtifolia, Populus

spp.; Jach & Ceulmans, 1999; Li et al., 2000; Sigurdsson,

2001; Tricker et al., 2004), no consistent pattern has

emerged yet, either within or among species or forest

types, reflecting perhaps the length of the time series

available. For example, elevated [CO2] affected neither

emergence nor abscission date of either sun or shade

leaves of canopy sweetgum at our site (Herrick &

Thomas, 2003), but in another sweetgum forest it

caused a significant variation in canopy duration (Nor-

by et al., 2003). Because the effect was inconsistent

among the 4 study years, the authors of the latter study

concluded that there was no effect on leaf dynamics.

This study shows that [CO2]-enrichment can decrease

the rate of late season hardwood leaf loss (Fig. 4d),

perhaps reflecting a better leaf carbon balance and a

lesser need for water than plants under ambient [CO2]

(Schafer et al., 2002, 2003). Regardless of treatment,

some of the interannual variation in the L dynamics

was caused by extreme events such as the ice storm of

December 2002, and hurricanes (Fig. 5). Yet, most of the

interannual variation was caused by drought (Fig. 6)

and, for the pine component, to the degree of canopy

closure. Since 1998, Lprodp was positively related to the

minimum Lp – occurring at the beginning of the grow-

ing season – and soil moisture during foliage expansion.

The rate of Llossp was also affected by soil moisture

availability during the growing season. Low soil moist-

ure depressed the maximum rate of Llossp by increasing

litterfall during nonpeak times (Fig. 4c). Overall, envir-

onmental drivers had the greatest impact on the

dynamics of leaf area index, whereas elevated [CO2]

had little impact.

Fig. 6 1996–2003 ratio of peak pine leaf area (Lpeakp) to winter

minimum pine leaf area (Lminp) (a) and pine winter minimum

leaf area-to-sapwood area Lminp/Asp (b) for ambient CO2, ele-

vated CO2, fertilization and CO2 in combination with fertiliza-

tion. Bars indicate 1 SE. Events likely to have contributed to the

observed pattern are noted. Reference line in (a) is the average

ratio across all years and reference line in (b) is at 0.17, an

independently established allometric value for loblolly pine

(Pataki et al., 1998b).

Fig. 7 Treatment level enhancement ratios of average func-

tional leaf area ðLÞ for elevated CO2, fertilization and elevated

CO2 in combination with fertilization, for pine (a) and canopy

(pine 1 hardwood) (b). The additive effect (elevated CO2 1 ferti-

lization) is indicated with a dotted line. Bars indicate 1 SE.

Functional leaf area for pine is derived from months when

monthly average temperature is 49 1C, and canopy leaf area

is derived from the average functional pine leaf area plus the

average leaf area of hardwoods during their foliated period.

Enhancement ratios have been corrected for pretreatment differ-

ences, such that the enhancement ratios should represent only

treatment-induced enhancements.



Treatment-average leaf area index

To describe the treatment effect on leaf area index we

focused on some of its most relevant metrics, such as

Lpeak, Lmin, Lprod (see Appendices A and B) and the

functional L. In contrast to other studies in closed

canopy forests (Hattenschwiler et al., 1997; Gielen

et al., 2003; Norby et al., 2003), we found that the

enhancement of leaf area by elevated [CO2] was sus-

tained, albeit at a lower than initial level, even after the

canopy closed in 1999. Functional Lp enhancement

under elevated [CO2] stabilized at �16% (�1%), with

climate events introducing some variability (Fig. 7a).

We observed very little impact of elevated [CO2]

through direct assessment of hardwood L. In addition,

environmental variables were not successful in describ-

ing annual variations of hardwood L. A major factor

likely preventing us from detecting differences in Lprodh

was the uneven distribution of hardwoods across plots,

the effect of which was not overcome even with appro-

priate pairing of plots for statistical testing. Combining

Lh with Lp into functional Lc (Figs 5c and 7b) shows

similar patterns to those observed with Lp alone, as

expected since pine dominated in most plots. Thus,

after 1999 (the period after the canopy reached quasi-

equilibrium) functional Lc showed a [CO2]-induced

enhancement of 14% (�1%).

Spatial variability of leaf area and [CO2]-inducedenhancement

We evaluated the effect of [CO2] on the spatial distribu-

tion of leaf area. We assessed the effect on the vertical

distribution of L through the canopy, and on the hor-

izontal distribution of L over the site using plot-specific

information on N availability.

How leaf area is distributed vertically within a forest

canopy is important in determining the light environ-

ment within and below the canopy (Stenberg et al., 1994;

Larsen & Kershaw, 1996). Elevated [CO2] did not

change the vertical distribution of L in three poplar

species (Populus alba, P. nigra and P. � euramericana;

Gielen et al., 2003). Similarly, in our study [CO2] did not

strongly affected vertical distribution of L, although the

pine leaf area tended to move upward relative to

ambient plots, and individual crown length increased

by �6%.

A recent synthesis of results from four FACE sites

evaluated the response of L to elevated [CO2] after

canopy closure (defined as L490% of the maximum

reached in natural stands of the same species in the

region; Norby et al., 2005). The [CO2]-induced L

enhancement in hardwood canopies was relatively

large for canopies composed of species with inherently

low Lpeak, and decreased for species with high Lpeak.

The Duke FACE pine fell in with the general pattern

formed by the hardwood species, on average showing a

moderate response to [CO2], in line with its relatively

low value of native Lpeak. However, our results demon-

strate that spatial variation in the response of

L to [CO2] can be caused not only when the control on

maximum L is the species composition, but also when

the availability of resources (e.g. of N) limits L.

This study was performed in a �0.5� 1 km stand,

uniformly planted on a relatively homogenous site with

level topography, yet there was readily discernable

horizontal variability in L (Oren et al., 2006). We found

that the CV of functional Lp in ambient [CO2] ranged

15–27% among years. Nitrogen availability also ranged

Fig. 8 Relative leaf area density (LAD) of pine as a function of relative height above the ground under ambient CO2, elevated CO2,

fertilization and elevated CO2 plus fertilization at four measurement times (a) day 320, 2002, (b) day 330, 2003, (c) day 66, 2003, (d) day 72,

2004. The fertilized treatments were measured only at day 66, 2003 and day 72, 2004.



considerably (Finzi et al., 2002), and already at the

beginning of the study significantly explained the var-

iation in L (Fig. 9a–d). With time, even more of the

spatial variation in Lp was explained by N availability

(Fig. 9a and c). Moreover, where N was low, the limited

development of pine canopy under ambient [CO2]

provided a greater opportunity for hardwoods to estab-

lish and develop crowns.

Following the commencement of [CO2] enrichment,

the increase in functional Lp was also controlled by N

availability, which explained much of the variability

among the [CO2] enriched plots (Fig. 9a and c). We

used the data from fertilized plots to assess whether the

[CO2]-induced responses observed under native N are

likely increase with further increase in N availability

(Fig. 9c). The data suggested that in contrast to the

forest under ambient [CO2], the maximum site Lp under

elevated [CO2] was achieved near the maximum native

available soil N (�5 g N m�2 yr�1). Thus, no further

increase in functional Lp is expected with further in-

crease in N through fertilization. Taken together, the

[CO2]-induced Lp enhancement ratio would increase

from zero over the range of native fertility, and decrease

thereafter as greater availability of N allowed Lp under

ambient [CO2] to approach the site maximum (Fig. 9c).

When hardwoods and pine were considered together,

there was little spatial variation in Lc, and it was not

controlled by N availability (Fig. 9b and d). Under both

ambient and elevated [CO2], the response of hardwood

L (mostly due to canopy individuals) at very low N

compensated for low pine L, and brought Lc to the site

maximum across the entire range in N. This response is

different from that observed in the pine, and resulted in

a constant [CO2]-induced enhancement ratio of Lc

throughout the entire range of native N. These results

suggests that on nutrient poor sites the pine may not be

able to respond to elevated [CO2], as was also shown for

wood production (Oren et al., 2001), but that certain

hardwood species may be able to respond. We note,

however, that lack of data disallows the evaluation of

the potential response of hardwood species to [CO2]

under more limiting N availability.

Given the rarity of FACE experiments, application of

their results to a meaningful scale must be done

through model extrapolation to regions and larger

areas. For these extrapolations, it is essential to quantify

interaction effects such as these described above. In

summary, this study demonstrates that the variation

in the leaf area of the pine at this site is not random, but

is strongly affected by plant available nitrogen and

climate. We show that, spatially, the response of pine

leaf area to elevated [CO2] is correlated with nitrogen

availability, but that this pattern disappears when total

canopy leaf area is evaluated because hardwood leaf

Fig. 9 Average functional leaf area ðLÞ as a function of the available N for ambient and elevated [CO2] in 1996 (pretreatment) and 2001

(year of maximum leaf area index), for pine (a, c) and canopy (pine 1 hardwood) (b, d). Regressions in (a, b) include only data from the

replicated FACE, while regressions in (c, d) also include data from the prototype complex. Average values from fertilized treatments are

included in (c, d) to suggest upper bounds on the response of L to elevated [CO2]. Canopy functional leaf area for 2001 is represented as

treatment averages due to the lack of a significant relationship with available N.



area compensates for low pine leaf area in low fertility

sections of the site (Fig. 9). That availability of different

resources interacts in affecting many growth processes

is not new. Allowing for the possibility of such interac-

tions would facilitate a greater understanding of the

mechanisms driving the responses of forests to elevated

[CO2]. The commencement of split-plot fertilization

within the replicated FACE experiment will permit a

formal analysis of some of these interactions.

Acknowledgements

We thank J. S. Pippen, A. Melvin, S. Gach, J. Janssen, J. Monfort,and J. Sibley for assistance with litter sorting and LAI-2000measurements. We also thank A. C. Oishi, and Drs S. Palmroth,R. H. Waring, and P. C. Stoy for useful comments. This study wassupported by the Department of Energy through the Office ofBiological and Environmental Research and its National Institutefor Global Environmental Change, Southeast Regional Center atthe University of Alabama, and by the US Forest Service throughboth the Southern Global Climate Change Program and theSouthern Research Station.

References

Albaugh TJ, Allen HL, Dougherty PM, Kress LW, King JS (1998)

Leaf area and above- and belowground growth responses of

loblolly pine to nutrient and water additions. Forest Science, 44,

317–328.

Augspurger CK, Bartlett EA (2003) Differences in leaf phenology

between juvenile and adult trees in a temperate deciduous

forest. Tree Physiology, 23, 517–525.

Brix H (1981) Effects of thinning and nitrogen fertilization on

branch and foliage production in Douglas-fir. Canadian Journal

of Forest Research, 11, 502–511.

Curtis PS, Wang X (1998) A meta-analysis of elevated CO2 effects

on woody plant mass, form, and physiology. Oecologia, 113,

299–313.

Davidson EA, Belk E, Boone RD (1998) Soil water content and

temperature as independent or confounded factors controlling

soil respiration in a temperate mixed hardwood forest. Global

Change Biology, 4, 217–227.

DeLucia EH, George K, Hamilton JG (2002) Radiation-use effi-

ciency of a forest exposed to elevated concentrations of atmo-

spheric carbon dioxide. Tree Physiology, 22, 1003–1010.

Dougherty PM, Whitehead D, Vose JM (1994) Environmental

influences on the phenology of pine. Ecological Bulletin, 43, 64–

75.

Drake BG, Gonzalez-Meler MA (1997) More efficient plants: a

consequence of rising atmospheric CO2? Annual Review of

Plant Physiology and Plant Molecular Biology, 48, 609–639.

Ducey M, Allen HL (2001) Nutrient supply and fertilization

efficiency in midrotation loblolly pine plantations: a modeling

analysis. Forest Science, 47, 96–102.

Ellsworth DS (1999) CO2 enrichment in a maturing pine forest:

are CO2 exchange and water status in the canopy affected?

Plant, Cell and Environment, 22, 461–472.

Ellsworth DS, Oren R, Huang C, Phillips N, Hendrey GR

(1995) Leaf and canopy responses to elevated CO2 in a

pine forest under free-air CO2 enrichment. Oecologia, 104,

139–146.

Ewers BE, Oren R, Sperry JS (2000) Influence of nutrient versus

water supply on hydraulic architecture and water balance in

Pinus taeda. Plant, Cell and Environment, 23, 1055–1066.

Finzi AC, DeLucia EH, Hamilton JG, Richter DD, Schlesinger

WH (2002) The nitrogen budget of a pine forest under free air

CO2 enrichment. Oecologia, 132, 567–578.

Finzi AC, Moore DJP, DeLucia EH et al. (2006) Progressive N

limitation of ecosystem processes under elevated CO2 in a

warm-temperature forest. Ecology, 87, 15–25.

Gielen B, Liberloo M, Bogaert J, Calfapietra C, DeAngelis P,

Miglietta F, Scarascia-Mugnozza G, Ceulmans R (2003) Three

years of free-air CO2 enrichment (POPFACE) only slightly

affect profiles of light and leaf characteristics in closed cano-

pies of Populus. Global Change Biology, 9, 1022–1037.

Gower ST, Norman JM (1991) Rapid estimation of leaf area

index in conifer and broad-leaf plantations. Ecology, 72,

1896–1900.

Hanson PJ, Wullschleger SD, Bohlman SA, Todd DE (1993)

Seasonal and topographic patterns of forest floor CO2 efflux

from an upland oak forest. Tree Physiology, 13, 1–15.

Hattenschwiler S, Miglietta F, Raschi A, Korner CH (1997)

Morphological adjustments of mature Quercus ilex trees to

elevated CO2. Acta Oecologia, 18, 361–365.

Hendrey GR, Ellsworth DS, Lewin KF, Nagy J (1999) A free-air

enrichment system for exposing tall vegetation to elevated

atmospheric CO2. Global Change Biology, 5, 293–309.

Hennessey TC, Dougherty PM, Lynch TB, Wittwer RF, Lorenzi

EM (2004) Long-term growth and ecophysiological responses

of a southeastern Oklahoma loblolly pine plantation to early

rotation thinning. Forest Ecology and Management, 192, 97–116.

Herrick JD, Thomas RB (2003) Leaf senescence and late-season net

photosynthesis of sun and shade leaves of overstory sweetgum

(Liquidambar styraciflua) grown in elevated and ambient carbon

dioxide concentrations. Tree Physiology, 23, 109–118.

Hofmockel KS, Schlesinger WH (2007) Carbon dioxide effects on

heterotrophic dinitrogen fixation in a temperate pine forest.

Soil Science Society of America Journal, 71, 140–144.

Hunter AF, Lechowicz MJ (1992) Predicting the timing of

budburst in temperate trees. journal of Applied Ecology, 29,

597–604.

Jach ME, Ceulmans R (1999) Effects of elevated atmospheric CO2

on phenology, growth and crown structure of Scots pine (Pinus

sylvestris) seedlings after two years of exposure in the field.

Tree Physiology, 19, 289–300.

Jarvis PG, Leverenz JW (1983) Productivity of temperate, decid-

uous and evergreen forests. In: Physiological plant ecology IV.

Encyclopedia of Plant Physiology, Vol. 12D (ed. Lange OL),

pp. 233–280. Springer-Verlag, New York.

Karnosky DF, Zak DR, Pregitzer KS et al. (2003) Tropospheric O3

moderates responses of temperate hardwood forests to ele-

vated CO2: a synthesis of molecular to ecosystem results from

the Aspen FACE project. Functional Ecology, 17, 289–304.

Kellomaki S, Wang KY (1997) Effects of long-term CO2 and

temperature elevation on crown nitrogen distribution and



daily photosynthetic performance of Scots pine. Forest Ecology

and Management, 99, 309–326.

Kinerson RS, Higginbotham KO, Chapman RC (1974) The dy-

namics of foliage distribution within a forest canopy. Journal of

Applied Ecology, 11, 347–353.

Larsen DR, Kershaw JA (1996) Influence of canopy structure

assumptions on predictions from Beer’s law. A comparison of

deterministic and stochastic similations. Agricultural and Forest

Meteorology, 81, 61–77.

Law BE, Van Tuyl S, Cescatti A, Baldocchi DD (2001) Estimation

of leaf area index in open-canopy ponderosa pine forests at

different successional stages and management regimes in

Oregon. Agricultural and Forest Meteorology, 108, 1–14.

Li J-H, Dijkstra P, Hymus GJ, Wheeler RM, Piastuch WC, Hinkle

CR, Drake BG (2000) Leaf senescence of Quercus myrtifolia as

affected by long-term CO2 enrichment in its native environ-

ment. Global Change Biology, 6, 727–733.

Lichter J, Lavine M, Mace KA, Richter DD, Schlesinger WH

(2000) Throughfall chemistry in a loblolly pine plantation

under elevated atmospheric CO2 concentrations. Biogeochem-

istry, 50, 73–93.

Linder S (1987) Canopy dynamics and growth of Pinus radiata. I.

Effects of irrigation and fertilization during a drought. Cana-

dian Journal of Forest Research, 17, 1157–1165.

Lui S, Teskey RO (1995) Responses of foliar gas exchange to long-

term elevated CO2 concentrations in mature loblolly pine

trees. Tree Physiology, 15, 351–359.

Maier CA, Johnsen KH, Butnor J, Kress LW, Anderson PH (2002)

Branch growth and gas exchange in 13-year-old loblolly

pine (Pinus taeda) trees in response to elevated carbon

dioxide concentration and fertilization. Tree Physiology, 22,

1093–1106.

Medlyn BE, Badeck F-W, De Pury DGG et al. (1999) Effects of

elevated [CO2] on photosynthesis in European forest species: a

meta-analysis of model parameters. Global Change Biology, 22,

1475–1495.

Medlyn BE, Barton CVM, Broadmeadow M et al. (2001) Stomatal

conductance of forest species after long-term exposure to elevated

CO2 concentration: a synthesis. New Phytologist, 149, 247–264.

Murthy R, Dougherty PM, Zarnoch SJ, Allen HL (1996) Effects of

carbon dioxide, fertilization and irrigation on photosynthetic

capacity of loblolly pine trees. Tree Physiology, 16, 537–546.

Naumberg E, Ellsworth DS (2002) Short-term light and leaf

photosynthetic dynamics affect estimates of daily understory

photosynthesis in four tree species. Tree Physiology, 22,

393–401.

Norby RJ, DeLucia EH, Gielen B et al. (2005) Forest response to

elevated CO2 is conserved across a broad range of productivity.

Proceedings of the National Academy of Science, 102, 18052–18056.

Norby RJ, Sholtis JD, Gunderson CA, Jawdy SS (2003) Leaf

dynamics of a deciduous forest canopy: no response to ele-

vated CO2. Oecologia, 136, 574–584.

Oren R, Ellsworth DS, Johnsen KH et al. (2001) Soil fertility limits

carbon sequestration by forest ecosystems in a CO2-enriched

atmosphere. Nature, 411, 469–472.

Oren R, Ewers BE, Todd P, Phillips N, Katul G (1998) Water

balance delineates the layer in which soil moisture affects

canopy conductance. Ecological Applications, 8, 990–1002.

Oren R, Hsieh C-I, Stoy P, Albertson J, McCarthy H, Harrell P,

Katul G (2006) Estimating the uncertainty in annual net

ecosystem carbon exchange: spatial variation in turbulent

fluxes and sampling errors in eddy-covariance measurements.

Global Change Biology, 12, 883–896.

Oren R, Pataki DE (2001) Transpiration in response to variation

in microclimate and soil moisture in southeastern deciduous

forests. Oecologia, 127, 549–559.

Palmroth S, Maier CA, McCarthy HR, Oishi AC, Kim H-K,

Johnsen KH, Katul GG, Oren R (2005) Contrasting responses

to drought of forest floor CO2 efflux in a Loblolly pine

plantation and a nearby Oak–Hickory forest. Global Change

Biology, 11, 421–434.

Palmroth S, Stenberg P, Smolander S, Voipio P, Smolander H

(2002) Fertilization has little effect on light-interception

efficiency of Picea abies shoots. Tree Physiology, 22, 1185–1192.

Pataki DE, Oren R, Phillips N (1998b) Responses of sap flux and

stomatal conductance of Pinus taeda L. trees to stepwise

reductions in leaf area. Journal of Experimental Botany, 49,

871–878.

Pataki DE, Oren R, Tissue DT (1998a) Elevated carbon dioxide

does not affect average canopy stomatal conductance of Pinus

taeda L. Oecologia, 117, 47–52.

Pearcy RW (1990) Sunflecks and photosynthesis in plant cano-

pies. Annual Review of Plant Physiology and Plant Molecular

Biology, 41, 421–453.

Raison RJ, Khanna PK, Benson ML, Myers BJ, McMurtrie RE,

Lang ARG (1992) Dynamics of Pinus radiata foliage in relation

to water and nitrogen stress: II. Needle loss and temporal

changes in total foliage mass. Forest Ecology and Management,

52, 159–178.

Rogers A, Ellsworth DS (2002) Photosynthetic acclimation of

Pinus taeda (loblolly pine) to long-term growth in elevated

pCO2 (FACE). Plant, Cell and Environment, 25, 851–858.

Sampson DA, Allen HL (1995) Direct and indirect estimates of

leaf area index (LAI) for lodgepole and loblolly pine stands.

Trees, 9, 119–122.

Schafer KVR, Oren R, Ellsworth DS, Lai C-T, Herrick JD, Finzi

AC, Richter DD, Katul GG (2003) Exposure to an enriched

CO2 atmosphere alters carbon assimilation and allocation in a

pine forest ecosystem. Global Change Biology, 9, 1378–1400.

Schafer KVR, Oren R, Lai C-T, Katul GG (2002) Hydrologic

balance in an intact temperature forest ecosystem under

ambient and elevated atmospheric CO2 concentration. Global

Change Biology, 8, 895–911.

Sigurdsson BD (2001) Elevated [CO2] and nutrient status mod-

ified leaf phenology and growth rhythm of young Populus

trichocarpa trees in a 3-year field study. Trees, 15, 403–413.

Sokal RR, Rohlf FJ (1995) Biometry. W.H. Freeman and Company,

New York.

Stenberg P (1996a) Simulations of the effects of shoot structure

and orientation on vertical gradients in intercepted light by

conifer canopies. Tree Physiology, 16, 99–108.

Stenberg P (1996b) Correcting LAI-2000 estimates for the clump-

ing of needles in shoots of conifers. Agricultural and Forest

Meteorology, 79, 1–8.

Stenberg P, Kuuluvainen T, Kellomaki S, Grace JC, Jokela EJ,

Gholz HL (1994) Crown structure, light interception and



productivity of pine trees and stands. Ecological Bulletin, 43,

20–34.

Stenberg P, Palmroth S, Bond BJ, Sprugel DG, Smolander H

(2001) Shoot structure and photosynthetic efficiency along the

light gradient in a Scots pine canopy. Tree Physiology, 21, 805–

814.

Stogsdill WR, Wittwer RF, Hennessey TC, Dougherty PM (1989)

Relationships between throughfall and stand density in a

Pinus taeda plantation. Forest Ecology and Management, 29,

105–113.

Sword Sayer MA, Goelz JCG, Chambers JL, Tang Z, Dean TJ,

Haywood JD, Leduc DJ (2004) Long-term trends in loblolly

pine productivity and stand characteristics in response to

thinning and fertilization in the West Gulf region. Forest

Ecology and Management, 192, 71–96.

Therezien M, Palmroth S, Brady R, Oren R (2007) Estimation of

light interception properties of conifer shoots by an improved

photographic method and a 3D model of shoot structure. Tree

Physiology, 27, 1375–1387.

Tissue DT, Thomas RB, Strain BR (1997) Atmospheric CO2

enrichment increases growth and photosynthesis in Pinus

taeda: a 4 year experiment in the field. Plant, Cell and Environ-

ment, 20, 1123–1134.

Tricker PJ, Calfapietra C, Kuzminsky E et al. (2004) Long-term

acclimation of leaf production, development, longevity and

quality following 3 yr exposure to free-air CO2 enrichment

during canopy closure in Populus. New Phytologist, 162, 413–

426.

Vose JM, Allen HL (1988) Leaf area, stemwood growth, and

nutrition relationships in loblolly pine. Forest Science, 34, 547–

563.

Vose JM, Dougherty PM, Long JN, Smith FW, Gholz HL, Curran

PJ (1994) Factors influencing the amount and distribution of

leaf area of pine stands. Ecological Bulletins, 43, 102–114.

Wang KY, Kellomaki S, Liatnen K (1995) Effects of needle-age,

long-term temperature and CO2 treatments on the photosynth-

esis of Scots pine. Tree Physiology, 15, 211–218.

Waring RH (1983) Estimating forest growth and efficiency in

relation to canopy leaf-area. Advances in Ecological Research, 13,

327–354.

Woodward FI (1990) Global change: translating plant ecophysio-

logical responses to ecosystems. Trends in Ecology and Evolu-

tion, 5, 308–311.

Zhang S, Allen HL (1996) Foliar nutrient dynamics of 11-year-old

loblolly pine (Pinus taeda) following nitrogen fertilization.

Canadian Journal of Forest Research, 26, 1426–1439.



Table A1 Treatment means for peak (Lpeakp) and minimum leaf area (Lminp), leaf area production (Lprodp) and annual average leaf

area ðLpÞ for pine

Ambient SE Elevated SE Fertilized

Elevated�fertilized Effect P-value

Lpeakp

1996 2.72a 0.42 2.67a 0.33

1997 2.59a 0.37 2.82a 0.31

1998 2.74a 0.41 3.18b 0.25

1999 3.32a 0.40 3.83b 0.29

2000 3.36a 0.38 3.89b 0.37

2001 3.82a 0.29 4.47b 0.39 4.89ab 5.35b CO2 0.029

2002 3.28a 0.26 3.88b 0.40 3.55a 3.93ab Year o0.001

2003 2.78a 0.20 3.28b 0.28 3.00ab 3.59b Year�CO2 0.391

Lminp

1996 1.77a 0.28 1.73a 0.22

1997 1.41a 0.22 1.43a 0.19

1998 1.71a 0.24 1.94a 0.23

1999 1.63a 0.24 1.95b 0.16

2000 1.87a 0.19 2.09a 0.17

2001 2.05a 0.16 2.39b 0.21 2.61ab 2.85ab CO2 0.045

2002 2.01a 0.15 2.32b 0.22 2.46ab 2.75b Year o0.001

2003 1.89a 0.16 2.20b 0.18 1.91ab 2.41ab Year�CO2 0.093

Lprodp

1996 1.57a 0.25 1.56a 0.23

1997 1.72a 0.23 1.95a 0.23

1998 1.69a 0.19 1.94b 0.16

1999 1.89a 0.21 2.10a 0.17

2000 1.97a 0.15 2.34b 0.29

2001 1.99a 0.15 2.32b 0.21 2.60ab 2.85b CO2 0.032

2002 1.70a 0.12 2.05b 0.20 2.13ab 2.17ab Year o0.001

2003 1.29a 0.07 1.55b 0.15 1.62ab 1.69b Year�CO2 0.517

Lp

1996 2.11a 0.32 2.08a 0.25

1997 2.00a 0.29 2.13a 0.25

1998 2.20a 0.32 2.52b 0.24

1999 2.38a 0.30 2.76b 0.21

2000 2.53a 0.28 2.88b 0.27

2001 2.80a 0.21 3.25b 0.27 3.50ab 3.83ab CO2 0.031

2002 2.75a 0.21 3.15b 0.31 3.11ab 3.44ab Year o0.001

2003 2.11a 0.16 2.48b 0.22 2.23ab 2.67ab Year�CO2 0.142

All values are m2 m�2. Superscript letters indicate statistical differences between treatments (Po0.05).

Bold values indicate o0.05.

Appendix A



Table B1 Treatment means for peak leaf area (Lpeakh) for hardwoods, and peak leaf area (Lpeakc), leaf area production (Lprodc) and

annual average leaf area ðLcÞ for the canopy

Ambient SE Elevated SE Fertilized Elevated� fertilized Effect P-value

Hardwood

Lpeakh

1996 1.29a 0.16 1.28a 0.27

1997 1.63a 0.15 1.65a 0.46

1998 2.06a 0.25 2.40a 0.61

1999 1.84a 0.18 2.16a 0.74

2000 1.90a 0.28 2.32a 0.74

2001 2.16a 0.33 2.38a 0.52 2.35a 3.01a CO2 0.347

2002 1.59a 0.20 1.79a 0.33 1.65a 1.77a Year 0.001

2003 1.63a 0.13 1.89a 0.33 1.96a 1.93a Year�CO2 0.224

Canopy

Lpeakc

1996 3.84a 0.40 3.81a 0.05

1997 4.01a 0.27 4.23a 0.13

1998 4.65a 0.18 5.43b 0.33

1999 5.08a 0.21 5.92b 0.48

2000 4.95a 0.09 5.92b 0.39

2001 5.87a 0.12 6.75b 0.38 7.03b 8.11b CO2 0.0641

2002 4.71a 0.21 5.49b 0.25 5.12ab 5.58ab Year o0.001

2003 4.18a 0.13 4.94b 0.18 4.74ab 5.31b Year�CO2 0.266

Lprodc

1996 2.89a 0.28 2.89a 0.07

1997 3.36a 0.17 3.62a 0.22

1998 3.69a 0.06 4.36b 0.47

1999 3.76a 0.06 4.29a 0.58

2000 3.88a 0.10 4.67b 0.50

2001 4.17a 0.20 4.72a 0.40 5.01a 5.93b CO2 0.116

2002 3.36a 0.16 3.86a 0.21 3.86ab 3.96ab Year o0.001

2003 2.93a 0.09 3.46a 0.23 3.60b 3.64b Year�CO2 0.145�Lc

1997 2.74a 0.25 2.88a 0.06

1998 3.22a 0.20 3.75b 0.08

1999 3.19a 0.21 3.74b 0.18

2000 3.38a 0.15 3.98b 0.13

2001 3.80a 0.10 4.41b 0.22 4.60b 5.21b CO2 0.041

2002 3.56a 0.15 4.08b 0.19 3.98ab 4.37b Year o0.001

2003 2.80a 0.12 3.29b 0.11 3.10ab 3.51b Year�CO2 0.148

All values are m2 m�2. Superscript letters indicate differences at Po0.05.

Bold values indicate o0.05.

Appendix B



Temporal dynamics and spatial variability in the enhancement of canopy leaf area under elevated atmospheric CO 2

Documents